Page 74 - AJWEP-v22i2

Ghosh and Prakasam

3.2. FR 4. Results and discussion
The FR model is a bivariate statistical method 4.1. Flood hazard conditioning factors
commonly used in hazard and disaster susceptibility 4.1.1. Drainage density
analysis. Its simplicity and effectiveness in predicting Drainage density refers to the length of the channels
flood likelihood make it a popular choice for estimating in a drainage basin divided by the basin’s total area.
32
hazard likelihood. The model calculates the likelihood It is considered one of the most important parameters
of future flood occurrence by analyzing the distribution of fluvial landscape evolution under the influence of
of past flood events across various factors. This model rivers. Consequently, there is a positive correlation
26
suggests that the importance of a specific factor in between drainage density and flooding. As a result,
predicting flood occurrence is directly corresponding to the likelihood of flooding increases with higher stream
the ratio of its class within the controlling factor. The network density. As shown in Figure 3, the drainage
33
FSI for a pixel is determined by summing the flood density of the study area has been categorized into five
frequencies of all factors associated with that pixel. classes based on its spatial pattern, which reveals that
27
The model estimates pixel-wise flood potential by all the drainage density classes have almost similar
pixel-wise FR of all factors, which is calculated as an pixel counts. A positive correlation was noticed
equation. between the drainage density classes and their FRs.
The very high drainage density class had the highest
/
FR Lc Af (VII) FR (47.93%) and vice versa. The floodplains of the
ij =
Lt At Barak River basin have high (0.3 – 0.4) to very high
The FR of the factor classes has been calculated using (>0.4) drainage density, which defines its higher
Equation VII, where Lc indicates the training flood susceptibility to flooding.
location in i class of flood conditioning factor j, Lt is
th
the total number of training flood points, Af is the area 4.1.2. Elevation
or pixel number in the i class of j flood conditioning Elevation is a primary determinant of flood vulnerability.
th
factor, and At is the total pixel count or area of the j Studies have shown that elevation is the most influential
factor. variable in flood occurrence, with lower elevations
correlating to a higher likelihood of inundation. The
23
3.3. Receiver operating characteristics curve analysis shows a considerable variation in elevation
The AUC is used to validate or examine the capability from mountains to floodplains of the Barak River basin.
of a model to predict the probability of occurrence Figure 3 represents the elevation map of the study area,
of hazards and disasters. The receiver operating where the entire Barak River basin is classified into five
characteristics curve (ROC) is the tradeoff between elevation zones. The highest proportion (33.04%) of the
false positive and accurate positive rates along the X study area topography has elevations <250 m, followed
and Y axes. Plotting the sensitivity, quantification by very high (>1.000), which represents 21.82% of the
28
of the successful and non-successful events, and study area. The very low elevated areas of the Barak
1-specificity on the abscissa and ordinates are the key River basin have 100% FRs, which indicates the flood
aspects of it. 29,30 The value of the AUC curve varies occurrence in the study area and is bounded in the low
between 0.5 and 1.0; a value between 0.9 and 1.0 elevated regions.
indicates excellent success or prediction rate, 0.8 – 0.9
indicates very good, 0.7 – 0.8 indicates good, 0.6 – 0.7 4.1.3. Slope
indicates moderate, and <0.6 indicates weak prediction The slope is a measure of how steeply a line or surface
rate of the model. 26,31 The AUC equation is presented in inclines. It is calculated as the change in height
Equation VIII, relative to the horizontal distance and expressed as a
percentage or angle. It significantly influences surface
TP TN runoff and water infiltration into the ground, which
AUC= (VIII)
PN makes it a crucial factor in studying and predicting
flood occurrence. Figure 3 shows the slope map of
27
where TP or true positive is the flood pixels and TN the Barak River basin, which has been categorized
or true negative is the non-flood pixels. P is the flood into five classes. The hilly surrounding regions have
point sum, and N is the number of non-flood points. high (20 – 25°) to very high (>25°) slopes, and the

Volume 22 Issue 2 (2025) 68 doi: 10.36922/AJWEP025040019